ABSTRACT
Editors
Maybe a good way to begin would be for you to tell us how your work with attribution started.
KelleyI was first stimulated with these kinds of ideas by John Thibaut and from his experiments with Henry Riecken and then Lloyd Strickland. John and I did those studies on the minimal social situations in which two persons control one another’s outcomes but have no knowledge of exactly how they affect each other (Kelley, Thibaut, Radloff, & Mundy, 1962). There was a little attributional problem in the middle of these studies, having to do with the kind of logical processes that tend to catch my attention. The problem was one of a person analyzing his effect on the other by doing something like a covariation analysis. They couldn’t both be doing this at once, or their mutual analytic efforts would interfere with either one finding out anything. Subsequently, a number of my students did follow-ups of the Thibaut work: Ken Ring did an extension of the Thibaut and Riecken studying (Ring, 1964), and Arie Kruglanski did a follow-up of Strickland (Kruglanski, 1970). They were just getting into doing these things in the mid-60s, a little before I wrote that paper for the Nebraska Symposium on Motivation (Kelley, 1967). I had had an interest in social perception all along.
JonesIn fact, you did your dissertation on first impressions in social perception.
372KelleyYes. In part I became interested in social perception problems because Kurt Lewin wasn’t dealing with social perception, yet he was always talking about the results of perception. It seemed to me that there needed to be an analysis of the processes involved. I remember that as being the rationale for my thesis.
EditorsWas Heider much of an influence?
KelleyVery much so. I should not have forgotten that. When I was at Minnesota at the Laboratory for Research in Social Relations, the whole group would occasionally read a book and really study and discuss it. We did this for Heider, so I read Heider more closely than I might have otherwise. I also was killing two birds with one stone, because I reviewed it for Contemporary Psychology in 1960. In my review of Heider, I emphasized the balance and the attribution ideas and distinguished between them. I am almost certain that Fritz, and I think he is right, regards that as overlooking vast chunks of the book. It’s a very rich work and undoubtedly has some other kinds of themes that can be drawn out.
EditorsDo you think you may have been responsible for separating those two notions more then Heider originally intended? Somewhere back in the history of social psychology, people started talking about balance and not bringing along with it the ideas involved in attribution.
KelleyIn some way, the two streams that are divided seem to be due to the separate treatment Heider gave them in his various papers.
JonesEspecially the one on phenomenal causality and the other on cognitive organization.
KelleyJust one other small point on that. Really the main exception to balance is an attributional effect. It’s when you expect to agree about something with someone you don’t like. That’s a very powerful effect, an exception to balance almost any way you think about it. It is an illustration of an assumed consensus effect where you and this other person, even though you don’t like him, are assumed to be looking at the same external stimulus and responding to it. The assumed agreement implies you are both responding to the same entity property. This exception is another kind of thing that got me interested in the middle 60s.
EditorsNed, would you like to tell us how you got interested in attribution?
JonesI think Jerry Bruner started my interest in person perception by referring me to Asch’s (1946) warm—cold study. It struck me as very intriguing because it got away from the accuracy problem that was inhibiting developments in perception – questions concerning who makes the best judge or what kind of person makes the most perceptible target. Then John Thibaut also put me onto Heider through the causality paper. John was thinking about the Thibaut and Riecken (1955) study when I was with him at Harvard in the early 50s. I was terribly intrigued by the causality notion, by the fact that people have to truncate the causal sequence somehow. At some fairly early point we have to stop and 373say, “He is responsible” or “It is responsible” or something like that. Let me give you an example.
When I was a 4th-year student in clinical psychology, we had a group therapy seminar run by Elvin Semrad at Boston State Hospital. Although it was supposed to be a seminar, there were only two ground rules. One was that somebody had to take notes every time, and the other was that we had to write a term paper. Everything else was up for grabs. It turned into an early encounter group. Semrad had been doing this for some years and was very skillful at bringing out hostilities between group members, turning them on himself, and then having us analyze the authority problem. The problem that kept arising in this group was, is there any such thing as legitimate hostility? Under what circumstances can you get mad at somebody and really justify it, especially if you are a clinical psychologist and concerned with the infinite regress of causation and the determinants of behavior? That sort of stuck in my craw. Many of the early things I did were an attempt to work out this problem. Under what circumstances is hostility ameliorated or mitigated by some knowledge of the hostile person or the situation under which that hostile remark was produced?
In an early, complex, and not entirely successful study (Jones, Hester, Farina, & Davis, 1959), we had a subject verbally attacked by someone who was either presented as well adjusted or maladjusted. We were attempting to see if a person being attacked actually felt better about the attack and therefore dealt with it better, knowing it came from a maladjusted person.
The Jones and Davis (1965) paper really came out of an attempt in the introductory social psychology classroom to talk about person perception in terms of choice. I had a little model I used to put on the board. I don’t know if the students ever took much away from it, but when Berkowitz invited me to do a chapter for Advances in Experimental Social Psychology, it was just a logical unfolding of all the problems involved in the simple-minded notion that you can tell what a person is like if he has choice when he behaves. I think the word “attribution” got in the subtitle of that article.
When Heider was at Duke in 1962 and 1963 he talked a lot about attribution and was concerned with extensions of the Heider and Simmel study of animated geometric figures. I don’t remember any special emphasis on the term in the book, but I later went back and saw that it was in the 1944 article; and I’m sure it was in the book, too. It seemed like a happy term to deal with a fundamental process in person perception.
I think I came to attribution from a narrower perspective than Hal, because I had been concerned with two paradigms. One was person perception; the other was self-presentation. I think at the outset I didn’t really see clearly the connections between these things; however, now I see them as very closely related if only because self-presentation provides the figural stimulus for person perception in the real world. Initially, my interest in ingratiation was quite separate from my interest in person perception, but that didn’t last very long. I remember 374being stunned and a little envious and delighted when Hal’s 1967 paper came out. It was a comprehensive statement which tied together a lot of things that I had not thought about as being tied together. The social comparison process, for example, I hadn’t really related to attribution in any way. The pictorial representation of an analysis of variance cube was very heuristic.
KelleyActually, I added on those pictures of the cube and the ANOVA analogue after I had written the paper. In some ways I am not happy that I did it. It was a thought-aiding device, not something I wanted to feature. I should have known that the analogue, being very concrete, would provide the label.
JonesI think it’s very important to have a label. In this case it integrates a set of variables that otherwise seem unrelated.
